Jump to content

Bo (ABRP)

Administrators
  • Content Count

    155
  • Joined

  • Last visited

  • Days Won

    15

Everything posted by Bo (ABRP)

  1. To verify and improve our models we need your feedback. There's many ways we could use help if you actually own one: Drive a plan and compare it to the actual battery used. Drive a plan with the browser active, and update your actual battery percentage in the browser. Contribute data via OBD or other methods. The best way to improve the data is to provide data directly from the car. Connecting your car not only improves the models, but allows you to follow up directly in the car while driving! We have several ways of doing that, but we need your help to figure out what will work with your car: An OBD reader can be used in concert with a custom app like LeafSpy, or a PID list and Torque Pro. If your manufacturer has an API to access data from the car we can set up access to that API, like we do for Tesla If you're familiar with either of these, contact me at jason@abetterrouteplanner.com and I'll help figure out what we need to do to connect your car! Thanks for providing feedback!
  2. To verify and improve our models we need your feedback. There's many ways we could use help if you actually own one: Drive a plan and compare it to the actual battery used. Drive a plan with the browser active, and update your actual battery percentage in the browser. Contribute data via OBD or other methods. The best way to improve the data is to provide data directly from the car. Connecting your car not only improves the models, but allows you to follow up directly in the car while driving! We have several ways of doing that, but we need your help to figure out what will work with your car: An OBD reader can be used in concert with a custom app like LeafSpy, or a PID list and Torque Pro. If your manufacturer has an API to access data from the car we can set up access to that API, like we do for Tesla If you're familiar with either of these, contact me at jason@abetterrouteplanner.com and I'll help figure out what we need to do to connect your car! Thanks for providing feedback!
  3. To verify and improve our models we need your feedback. There's many ways we could use help if you actually own one: Drive a plan and compare it to the actual battery used. Drive a plan with the browser active, and update your actual battery percentage in the browser. Contribute data via OBD or other methods. The best way to improve the data is to provide data directly from the car. Connecting your car not only improves the models, but allows you to follow up directly in the car while driving! We have several ways of doing that, but we need your help to figure out what will work with your car: An OBD reader can be used in concert with a custom app like LeafSpy, or a PID list and Torque Pro. If your manufacturer has an API to access data from the car we can set up access to that API, like we do for Tesla If you're familiar with either of these, contact me at jason@abetterrouteplanner.com and I'll help figure out what we need to do to connect your car! Thanks for providing feedback!
  4. Bo (ABRP)

    Waypoint charging

    Thanks for the report! There was a bug which made waypoint charging return an error, just as you pointed out. However, for the other example with only time and percentage, this is actually not supported. You need to provide power (unless the waypoint is a known charger) and time or final SoC. This is not completely obvious from the user interface 🙂 Edit: Here is your route as I think you intended it: https://abetterrouteplanner.com/?plan_uuid=9e4a21e0-2fc9-471d-8802-c8791c4098b0
  5. New feature! ABRP now keeps track of charging networks/operators and allows you to prefer one or more of them. This is done by clicking a charger which has an operator and selecting "Prefer...". The planner will treat preferred chargers as "less costly" than other chargers will therefore more often use them. It is still a soft decision though, so if a non-preferred charger gives a significantly faster plan, it will still be selected. Let us know hownit works!
  6. ... and fixed. Thanks for the re-report of the same bug
  7. This is actually not a bug with OSRM, it is us. We don't have 3D coordinates for chargers, and this one is just on top of a highway lane. Last time I asked supercharge.info to move the coordinates 20 m south, and this solved the issue. Since then we have switched to using tesla.com directly instead for SuC positions, and they have the old/true position. I'll need to implement some position override, should be easy.
  8. This got lost when we transitioned from Google Maps. Now back again!
  9. I'd love to break it down more, but the issue is that Model 3s all report exactly the same option codes through the API, so I can only use what the owners select as car model. We recently included specific AWD 18- and 19-inch wheel models so hopefully this will enable us to look at the differences between AWD and non-AWD too. We do log outside temperature; we'll look at that once winter is coming (in the northern hemisphere). Otherwise, the median filtering we do at each speed means that the curve we get is for the most common weather and driving - likely nice dry summer driving since the data is mostly from the summer and early fall.
  10. ABRP finds a route by optimizing the total time, charging+driving+overhead. Getting to a charger in a modern Tesla with less than 10% is usually not a good idea, timewise, so ABRP will typically avoid that. There is no cost to getting to the destination at very low SoC %, so it will always arrive with the lowest allowed SoC. The ABRP optimizer is currently slightly granular and will typically optimize the time in steps of 5% charging - which is why it sometimes jumps from 10% to 15% arrival SoC if that turns out to be (slightly) better.
  11. The issue with Model 3 specifically is that the option codes reported by the MyTesla API are broken - they are all the same for any Model 3. This means that we have to trust what the user chooses in ABRP as car model, and this is quite often not completely correct. Also, we have, so far been lacking a choice for AWD. Will fix that soon!
  12. Thanks for the catch, the imperial unit efficiency numbers were switched. Now fixed! Also, we do not treat 110 km/h as equal to 65 mph. These speeds were chosen to be common highway speeds in the different regions. But perhaps we should have chosen 120 km/h which is pretty much 75 mph spot on 🙂
  13. As you may already know, ABetterRouteplanner.com collects driving data from contributing users. This data is used to improve the ABRP car models (i.e. the mathematical representation of each type of car) and also, to give back to the ABRP community providing the data, to be published here in the blog! At this point, we have received a lot of Model 3 Long Range driving data, and even enough data from Model 3 Performance to draw some first conclusions on how they differ. We have 220 different Model 3 Long Range users who have contributed a total of 70,000 km (43,000 miles) of driving, which caters for very good data. Most of these Model 3 Long Range are the RWD version. For P3D, there is "only" 13 cars contributing so far, having driven around 7,000 km (4,300 miles) of driving, which means that statistics is a little bit more shaky, but still usable. Just to show off what those numbers mean, this is the graph of the 7,000 km of P3D data - every blue point corresponds to 30 seconds of driving at a certain speed and how much power that car has consumed during those 30 seconds. Yellow dots show the median power consumption for that speed and red line is our fitted model. Now let's boil that data down to something more understandable! First, the reference consumption, constant speed on flat land at 110 km/h (65 mph) becomes: Tesla P3D: 173 Wh/km at 110 km/h (267 Wh/mile at 65 mph) Tesla Model 3 Long Range, mostly RWD: 150 Wh/km at 110 km/h (232 Wh/mile at 65 mph) This means that our real-world driving data shows that the P3D consumes about 15% more than the RWD version at highway speeds. This is expected, or actually somewhat low - most P3Ds run on 20" sports wheels instead of 18" Aeros, and perhaps more importantly, P3Ds may be driven more like performance cars by their drivers. (Note that in https://abetterrouteplanner.com, we add some margin to the reference consumption to be on the safe side.) Looking at efficiency numbers for different speeds, we get the following comparison between the Tesla Model 3 Long Range, P3D and Model S100D: We can see from this graph that there is a clear difference in efficiency between the RWD and the P3D, and at higher speeds where the Aero wheels with better aerodynamics, the difference grows even more. This efficiency leads to this range-vs-speed graph: And finally, in our standard road-trip challenge of a virtual 1,000 km (621 mile) drive with fast chargers every 200 km gives us: Tesla Model 3 Long Range: Total trip time 09:44, of which charging 01:24. Tesla Model P3D: Total trip time 09:59 of which charging 01:39. Tesla Model S100D: Total trip time 10:05 of which charging 01:45. So the slightly higher consumption in a P3D does equate to 15 minutes more charging time in a 1,000 km road trip. Not too shabby! Appendix - Graphs in Imperial Units
  14. Driving mode for Everyone! A Better Routeplanner has always been about two things to make life as an EV owner as smooth as possible: Planning at home, and following up while driving. In a Tesla, specifically with telemetry via MyTesla, the ABRP driving mode has been working smoothly, but in a mobile phone or iPad, driving mode has been absent. Now ABRP 3.3 changes that! Release highlights Driving mode, with or without telemetry. As soon as you move fast enough, ABRP will switch to driving mode and display a graph of the next route leg including elevation and expected battery State-of-Charge % (SoC) together with estimated arrival time. Full screen mobile web app. If you go to ABRP in your mobile phone or iPad browser, you can “Add to Home Screen” in your browser. The ABRP icon on your home screen will then launch ABRP in a full screen mode which works more like an ordinary app. More space for graphs! Avoid ferries, tolls and highways. If you want to have more interesting routes, you can now disable any of these. This also improves the previous avoid ferries function which was not ideal. OBD telemetry. Just like we fetch SoC and other car data via MyTesla, we can use OBD data for other car brands and get real-time SoC and other information for convenience while driving. How to set up an OBD reader with ABRP. Manual SoC input. If you are not fetching information from your car via telemetry (MyTesla or ODB), you can now manually input your actual SoC. This is shown as a big battery symbol above the driving mode window. Improved reference consumption estimation. With telemetry or manual battery input, ABRP will now estimate your reference consumption, even if you do not have a route planned. The end user’s license agreement has been updated to allow ABRP to store anonymous drive data. There is no way to track this data back to any person, even if our servers get hacked, and we will use it to improve our planning and car models even further for a better experience for everyone. The lowlight of the release is Teslas v9 software, which forces their own map in the background at all times. For Model S and X users with the old MCU (before March 2018) this makes ABRP almost unusably slow in the car browser. Tell your Tesla representatives. Driving Mode with Manual SoC Input If you are driving a car without telemetry, you can still get a visual indication of your actual SoC (battery %) compared to the plan by letting ABRP know your actual SoC. Do this by adjusting the estimated SoC in the green battery icon to match your car’s SoC. Click to the left of the icon to decrease the actual SoC Click to the right of the icon to increase the actual SoC Click in the middle of the icon to confirm that ABRP’s estimate is correct By doing this you get a blue SoC graph overlaid on the grey planned SoC in driving mode and you can visually see how you are doing. Also, you help ABRP improve our car models and planning, and thereby help all other fellow EV owners!
  15. Introduction One of the questions we get fairly regularly at ABRP is if and when we will support other Electric Vehicles in the planner. To create a model the planner can use we need two aspects defined: Driving Consumption and Charging. Once we have a model for a car, we have three markers we’ll put in the planner: Alpha – An initial model based on measurements or data by external parties, not thoroughly verified. Beta – A more mature model validated by matching to owners’ actual road trip numbers Release – Model verified and improved by recorded real life driving data. Once you have the data we need to make Driving Consumption and Charging Models, email it to the ABRP Team, and we’ll work with you to get the model on the site. We won’t make any car model available on ABRP without first verifying accuracy with an actual owner of a car, so it helps immensely if you’ve got an example road trip you can recreate in the planner once we’ve prepared the model. Emails: bo@abetterrouteplanner.com jason@abetterrouteplanner.com Driving Consumption Model The driving model is the most important part of the planner, as it calculates how much battery you’ll use on each leg of the trip. To do this calculation, the planner uses a third order polynomial to calculate consumption. This is based on the physics of driving. There are several ways to create this driving model, listed here in order of accuracy: 1 – Analytical Physics Model Creating an analytical model requires the least access to a car, but it’s the least guaranteed to be accurate. To create an analytical model, we use the physics of driving: Pdrag = η*Cd*A*v3 Prolling resistance = η*Crr*M*g*v Pidle = Constant Combining these three terms, we get a full equation for the consumption of a given car. We then multiply by the drivetrain efficiency, what percent of battery energy goes into creating motion. To create this model, we need to fill in a few variables: Parameter Description Cd Coefficient of Drag - Typically available online, but takes a little bit of research to verify. A Frontal Area - Cross-sectional area of the vehicle. This can be calculated roughly from front or rear photos of the vehicle and the vehicle's dimensions Crr Coefficient of Rolling Resistance - Mainly based on the tires, and drivetrain of the vehicle. A little harder to dig up. This typically varies from about 0.007 for very efficient tires to 0.014 for wide performance tires. M Mass - Curb mass of the vehicle. Pidle Idle Power - The amount of power drawn by the vehicle when sitting still. Best drawn by looking at the built-in energy meter while at a stop. Typically around 1-1.5kW η Drivetrain Efficiency - Efficiency of turning battery power into movement. Generally between 85%-95% If you can provide all of these parameters, we can create a model that is fairly accurate. 2 – Manually Collected Driving Data The next-best method is calculating the driving model from approximate real-world data. This can be gathered by recording consumption while driving. By observing the power draw at various speeds on flat ground, and providing data points we can calculate a better consumption model. In general, we need the highest number of data points at freeway speeds, as you tend to spend most of your driving on a trip at those speeds. I would recommend noting the power consumption readout at the following speeds (km/h and mph are not exactly equivalent, for ease of use, please note which unit you used to collect data so our model can be as accurate as possible). If you plan to use this method, be sure to drive safely, you may want to recruit an assistant to take the power draw notes while you drive. Speed (km/h) Speed (mph) Power (kW) 30 20 45 30 60 40 75 50 90 55 100 60 110 65 120 70 130 75 140 80 150 85 3 – Collecting Driving Data Directly This is the best way to build an accurate model. This can vary by vehicle or manufacturer, but most vehicles provide data via the OBD port. For an example, see the for the generic OBD and Torque instructions. To contribute data this way, set yourself up with Torque Pro, and find a PID list for your car. If you can find these things, and can verify the data shown on the Realtime Information display in Torque is accurate, contact me (jason@abetterrouteplanner.com) to set up the server to receive your data. Once there’s enough data for your vehicle, we’ll perform the same analysis we’ve done for all the currently released models in the planner. Charging Model The charging model is a little easier to build from available information. If you can find a plot of the maximum charging speed relative to State of Charge, or a video that shows power in kW and battery %, we can build a charging model from that. Something like the following chart (sourced from Fastned for the Hyundai Ioniq): Fastned charging data for the Hyundai Ioniq Again, however, the best way to build a charging model is to contribute OBD data directly from a charging session. See the setup instructions in the previous section for what’s needed to submit OBD data.
  16. Thanks to the 5 Bolt EV drivers contributing driving data to ABRP, we’ve finally got enough data points to define our first real world consumption curve for the Chevy Bolt EV! We can still use driving data to improve the model even further, so if you’d like to contribute your data, have a look at the instructions. Also, if you’ve got an electric car you’d like us to support, contact myself and Bo, and we’ll run you through what we need to add the car to the planner. Much of the data comes from day-to-day driving, with a few road trips sprinkled. Thanks to everyone who contributed their driving data! Bolt EV Power at Constant Speed The ABRP Model, as you know, is driven by calculating the power consumed for each leg of the trip by plugging in the speed, elevation, and other factors for each segment of the trip, determining how much power the car must output to drive that segment. It then adds up all the segments, and subtracts it from the battery to determine when a charging stop is needed. For this purpose, we need an accurate driving model on flat ground as a baseline, that turns speeds into power. Then we can add all the other driving factors on top of that: Bolt driving data fit to a third order polynomial The blue dots are consumption samples (30 seconds of driving), adjusted for elevation and speed changes, and the yellow dots are median points within a 1 m/s bin. This means that the red line consumption model is fitted to the median consumption which means that most weather issues, car defects, aggressive driving and so on is typically ignored. As can be seen, we mostly have data points from very low speed and from highway speeds, but that is where things are most interesting. For the Bolt EV, the bottom line is, it’s actually a quite efficient vehicle, though its somewhat unaerodynamic shape really hits it at high speeds. The model gives us a reference efficiency of: 6.06 km/kWh (165 Wh/km) at 110 km/h or 3.92 mi/kWh (255 Wh/mile) at 65 mph That’s pretty good for a boxy little hatchback! For comparison, that’s about halfway between the Model 3 (143 Wh/km) and the Model S (188 Wh/km). We’ve now updated the live model for the Chevy Bolt (and Ampera-E), so you should see the benefits of this higher efficiency in your route planning! Do note that we still set the default a little lower than that, just to ensure we give you a plan that’s not going to over-promise your car’s capabilities. Comparing to the Analytical Model Driving range vs speed comparison between the real world driving data and the original analytical model. Up to this point, we’ve been using an analytical model, using the drag characteristics, rolling resistance, and other parameters to determine an approximate driving model for the Bolt. Since we’ll be building a lot of these as more EVs come to market, we wondered how accurate the analytical model really is: As you can see, the models match quite closely! In fact, when building the Bolt analytical model, I added a 10% margin of safety to the model until we could validate using real world data, and you can see that at freeway speeds, the Analytical Bolt is about 10% lower than the Real World Bolt. Road Tripping in the Bolt With all the data we’ve got, let’s add the Bolt into the road table, and see where it falls: Model 3 Long Range: Total trip duration 09:43, of which charging 01:23 Model S100D: Total trip duration 10:05, of which charging 01:45 Model X100D: Total trip duration 10:29, of which charging 02:09 Model S60: Total trip duration 10:35, of which charging 02:15 Model X60: Total trip duration 11:28, of which charging 03:08 Bolt EV: Total trip duration 12:20, of which charging 04:00 This is using the “ABRP hypothetical road trip” is 1000 km (621 mi) in 200 km (124 mi) steps at 120km/h (75mph). The Bolt’s overall road trip speed is pretty slow. Even slower than the slowest Tesla. This is a consequence of the relatively slow charging that Chevy has built into the Bolt: Charge speed comparison, accounting for vehicle efficiency and battery size. Comparing the Bolt’s charge speed to that of the Model 3, we can see why our road trip takes so long! Accounting for battery size and driving efficiency differences, the Model 3 can charge nearly 2.5 times faster than the Bolt at each relative peak. All in all, the Bolt definitely can do those road trips, but it’s going to be at a much slower overall pace than the Model 3. The upcoming Hyundai Kona is a little bit faster than the Bolt, but not hugely, since it’s slightly less efficient. The larger battery makes for slightly longer range, but also means it takes a little longer to charge. Once we start getting some data, we’ll do a comparison to see how much faster the Kona really is than the Bolt. Appendix: Graphs in Imperial Units
  17. We value actual driving and charging data very highly at ABRP. We use it both to research and publish information about the efficiency and charging characteristics of Electric Vehicles, and to power our internal models of the cars for trip planning. There are several posts on this blog about the charging and consumption characteristics of Tesla Model S, X, and 3. And all of this is made possible thanks to our generous users who donate their data! Tesla vehicles are, of course, all online all the time. They also have an inofficial API which we can use in ABRP to fetch information e.g. the current State-of-Charge (SoC, battery %) from your car. All other present EVs are either not always connected to the internet, or they lack an open API which we can use to fetch information. To obtain data from these EVs we rely on the standardized OBD port in every vehicle – an example can be seen in our post on submitting Bolt data. Hopefully this will be easier and more accessible to everyone at some point. Tesla Data Sharing If you drive a Tesla, you can easily help out and submit your driving and charging data to ABRP. Just browse to https://abetterrouteplanner.com in your car browser or mobile/computer and select Settings->More Settings->MyTesla login and login using your MyTesla email and password. There are two options below the login details: Share data: Check this if you want to donate all data your browser fetches for you while driving, planning and charging. We will only poll data from your vehicle when you have an active browser window running ABRP. 24h background sharing: If you want to share data, but not have a browser running ABRP all the time, check this option too. It will allow the ABRP servers to poll your vehicle for up to 24 hours after you use ABRP. For e.g. Model 3s, which still lack a browser, this helps a lot if you want to donate quality data from your trips. Just visit ABRP once in a browser in the beginning (or before) the trip. If your car is idle, polling will only happen once per hour to limit vampire drain. Access Security We take security very seriously. We don’t want to store any sensitive information on our servers, and definitively not MyTesla tokens. We never ever store your MyTesla password anywhere. The MyTesla token used to access your vehicles API is only stored permanently in your browser’s local storage. The 24h background sharing feature needs your MyTesla token on our servers to work. This is solved by remembering it in the local memory of a server process when your browser asks for MyTesla data. It is never stored on disk or in a database, and will be forgotten after 24 hours or as soon as the server is rebooted. And of course, the feature is entirely optional, just like data sharing and logging in to MyTesla at all.
  18. Finally! Enough Model 3 owners have gotten through the hassle of keeping ABRP running in a mobile browser while driving and charging to give ABRP some initial consumption and driving statistics. Have you noticed how much better ABRP has become in the mobile browser with the latest UI updates, BTW? We have data points from 57 Model 3s, but most vehicles have only provided a couple of seconds of data. Most of the data actually comes from a Swedish American cross-east-cost driver who left ABRP running often enough in the phone – great thanks Pontus! The total Model 3 distance driven with ABRP running is only so far 1200 km, consuming 187 kWh, so more data is definitively needed. However, the data we have is pretty consistent and therefore we choose to publish it here. Model 3 Power Consumption at Constant Speed Let’s dive into data. Here goes the standard power-vs-speed chart that we rely on a lot at ABRP. This is data corresponding to driving at a constant speed on flat land. Elevation and speed changes are then added on top by ABRP when planning routes. The blue dots are consumption samples (30 seconds of driving), adjusted for elevation and speed changes, and the yellow dots are median points within a 1 m/s bin. This means that the red line consumption model is fitted to the median consumption which means that most weather issues, car defects, aggressive driving and so on is typically ignored. As can be seen, we mostly have data points from very low speed and from highway speeds, but that is where things are most interesting. The bottom line; ABRP reference consumption is impressively low at these summer temperatures: 143 Wh/km at 110 km/h or 218 Wh/mile at 65 mph That is very very good and a lot of credit has to be given to Tesla for working so hard on efficiency even at high highway speeds. Model 3 Charging Power While consumption while driving is very important for a good long trip experience with an EV, this is equally true for charging. Tesla’s supercharger network is fantastic, and one of the most important factors in making a Tesla the only vehicle you need. The Tesla supercharger can currently deliver up to 120 kW to one vehicle, but that is not the only limitation – the battery itself limits the power depending on its current State-of-Charge (SoC), i.e. battery %. ABRP has received charging data from 40 Model 3s at 185 charging sessions charging in total 1775 kWh (obviously it is more appealing to play with ABRP while charging the Model 3 than while driving, due to the lack of a browser in the car). The charging power curve of the Model 3 battery “BT37” is quite impressive: It basically takes as much power as the S/X 100D BTX6 battery at the same SoC %, but with a smaller battery capacity. The estimated usable capacity of the BT37 battery from this data is 73.4 kWh. Comparing Tesla Model S3X at Summer Temperatures In an attempt to compare the characteristics of all of Tesla’s present models we have plotted the consumption and range of a Model S100D, Model 3 Long Range and Model X100D based on ABRPs models – in turn based on all generously donated data. The data speaks for itself; X is obviously the most power hungry vehicle of the three, and the consumption model indicates that its is particularly bad at high speeds – but that is hardly a surprise given the sheer size of the X. Model 3 is indeed impressively effective. Even more impressive is that while Model S100D keeps the lead in the range league, Model 3 Long Range is not far behind even though it only has 75% of the battery size of the 100D: One more thing. The combination of low power consumption and high charging speed gives the Model 3 Long Range a unique position in the perhaps most interesting challenge: Shortest road trip time. Take a fictitious road trip with 200 km (124 miles) between Superchargers. Drive at a constant highway speed, 120 km/h (75 mph) between chargers for a total of 1000 km (620 miles). The result according to ABRP looks like this: Model 3 Long Range: Total trip duration 09:43, of which charging 01:23 Model S100D: Total trip duration 10:05, of which charging 01:45 Model X100D: Total trip duration 10:29, of which charging 02:09 So the overall winner in that road trip race is clearly Model 3 Long Range. Appendix – Graphs with Imperial Units
  19. In the previous posts about Tesla Model S power consumption vs speed (and the same thing for Model X) we dove through the surface of the large pool of driving data generously donated by ABRP users. There is more than 2.4 million driving data points now, and in this post we dive a little deeper. EVs and how they are affected by outside temperatures is usually a hot topic, but perhaps most during the winter. Well, now ABRP has collected data all through the (northern hemisphere) winter and the beginning of the summer, so we have enough data to compare consumption of Tesla Model S and Model X in different temperature ranges – so why wait? (Model 3 owners still have not provided enough data yet, unfortunately! Can’t wait for a browser in the Model 3.) Note that this is completely empirical data based on what is reported by the Tesla API from each vehicle, including power consumption, speed, and outside temperature. No guesses. Tesla Model S We have data from 511 vehicles with ABRP running while driving (you can donate data too by logging in to MyTesla in ABRP and leaving “Share Data with ABRP” checked). The method used to determine a power-vs-speed curve is the same as in the previous posts, and is based on finding the median steady-speed power consumption for each speed bin and then fitting a fourth order polynomial to the data. This polynomial can model everything from constant power consumption (even with zero speed) up to second order forces like drag (which become third order power). We split the driving data into four outside temperature bins: Below 0ºC (32F), between 0ºC and 10ºC (32F to 50F), between 10ºC and 20ºC (50F to 68F), and above 20ºC (68 F). First off, let us look at the consumed power-vs-speed for the four temperature ranges (if you prefer imperial units, have a look at the end of the post!): Obviously outside temperature matters. There is a relatively constant power gap between the highest temperatures and lowest temperatues, indicating that the heaters do consume something like 5-6 kW at the really low temperatures. But then of course, colder temperatures also correlate with generally worse weather, like snow and rain on the ground. So it is hard to be completely conclusive about the cause – but the data speaks for itself. Cold weather driving generally consumes a lot more power. Here is a different perspective on the same data; the maximum range of a Model S100D at different temperatures and speeds: Tesla Model X 294 Tesla Model X owners from all over the (western) world have donated data to ABRP. This data has been treated just like the Model S data above, with different power-vs-speed models fitted to data from four different temperature regions. Again, there is a clear difference between warm and cold outside temperatures when it comes to consumed power. (And again, imperial unit plots are available at the end of the post.) And the difference in power leads to different maximum range for a Tesla Model X100: Conclusions While it is unclear exactly what the source of higher consumption at lower temperatures is (heating, of course, but also worse ground conditions), it is very clear that on average, lower temperatures lead to significantly higher power consumption and therefore also less range. The maximum range of a Tesla Model S at temperatures below 0°C (32F) is about 22% less than at temperatures above 20°C (68F) – according to our real-world data from 511 vehicles. Pretty much exactly the same reduction, 22%, happens for the Model X. How Model 3 is affected by cold temperatures remains to be seen! Appendix: Plots with imperial units
  20. Similar to the post about Model S Consumption vs Speed, there is a lot of data from Model X owners driving around with ABRP active in their browsers and logged in with MyTesla. This is data from 1.2 million driving points from 279 Model X vehicles covering 160 000 km (100 000 miles), and almost all data points are included and corrected for elevation changes: The blue points are individual power consumption samples. The yellow points are the median power consumption for each speed (to remove outliers) and the red line is a fitted fourth-order polynomial model of power-vs-speed. At 30 m/s (approx. 110 km/h or 65 mph), the median consumed power for a Model X is about 25 kW. The ABRP reference consumption from this data becomes: Metric: 237 Wh/km at 110 km/h Imperial: 367 Wh/mile at 65 mph which is lower than the default settings in ABRP. Not that this is a complete mix of all factors such as different vehicles, wheels, temperatures, weather and so on. For fun, we can compare Model S and Model X curves: Model X consumption We can see that at low speeds, for example 10 m/s (36 km/h or 22 mph) the power consumption for both vehicles is pretty similar at about 7-8 kW. At high speeds, though, the much worse aerodynamic drag of the Model X comes in to play with full force. At 40 m/s (145 km/h or 90 mph), the Model S consumes 36 kW (250 Wh/km or 402 Wh/mile) whereas Model X needs 45 kW (312 Wh/km or 503 Wh/mile) to maintain speed. Note that you can easily convert from instantaneous power (in W) to energy-per-distance using the formula Energy-per-distance [Wh/distance] = Power [W] / speed [distance/h] Like this kind of data? Contribute data from your car too by logging in to MyTesla in ABRP and allowing data to be shared.
  21. Finally I have managed to get automatic refreshing of tokens to MyTesla to work in ABRP. In plain English this means that once you are logged in to MyTesla in ABRP, you should never have to log in again – it should not expire. Note that ABRP still does not store your password or MyTesla email anywhere. Only a temporary key from Tesla, called a token, is stored locally in your browser. Never on our servers. The refresh is done using Teslas refresh token, which is issued when logging in the first time. You can of course log out manually any time you like. Enjoy!
  22. Different batteries have different characteristics when it comes to charging; this goes for the various Tesla battery models too. Through the ABRP data collection by generous users, we have quite a lot of real-world data to base our models for use in the route planning, and in this post we give you some insight into the data. The data here is based on 4600 Supercharging sessions from 801 Tesla Vehicles! First of all, the battery model of a Tesla is not completely clear from the model name. An almost complete list of Tesla batteries includes: BT37: The 75 kWh battery in a Model 3 Long Range BT60: The old S60 60 kWh battery BT70: The old S70 70 kWh battery BT85: The classic “85” kWh battery in a Model S85 BTX4: The 90 kWh battery in S90 and X90 BTX5: The 75 kWh battery in S75 and X75 BTX6: The top-of-the-line battery 100 kWh in S100 and X100 BTX7: A rare 85 kWh battery, where we have almost no data BTX8: An 85 kWh battery found in some rare S75 and X75 Model 3 LR – BT37 First out is the Model 3 Long Range battery. There is a limited amount of data in ABRP’s database – only 38 charging sessions from 13 cars – so please contribute! The blue dots are measured data points and the red dashed line is the present ABRP charging power model. The estimated battery capacity from the contributing cars is 72.8 kWh Model S85 – BT85 Second, the classic S85 battery, which is known for not really being 85 kWh. There is plenty of data here, and as you can see it is charging at very high speed all the way from 0% SoC, but tapers off relatively early. The estimated battery capacity from the contributing cars is 73.4 kWh. That’s why I wrote “85”. It has basically the same capacity as an S/X75. Model S70 – BT70 The older S70(d) battery is similar to the BT85, but smaller. From data, the estimated usable capacity is 65.7 kWh. Model S/X90 – BTX4 The S/X90 battery, is like the “85 kWh” battery also not really living up to its name. The estimated capacity from the data is 79.8 kWh. It differs from the BT85 in that it charges slower at really low SoC (below 10%) but it compensates by charging a lot faster at higher SoC. Charging at BTX4 battery from 10 kWh to 50 kWh takes 23 minutes. The same charge (in absolute energy, not %) takes 27 minutes in a BT85. Model S/X75 – BTX5 The “new” 75 kWh battery, sometimes software limited to 60 kWh in a S/X60 has an estimated capacity of 71.6 kWh. The charging curve is similar to the BTX4 and BTX6 batteries, but in absolute power lower due to the smaller capacity. Charging from 10 kWh to 50 kWh takes 27 minutes. Model S/X100 – BTX6 The (so far) largest Tesla battery is a real beast. The charging is, in a large SoC region, limited by the 120 kW power output of most superchargers. 20 minutes to charge from 10 kWh to 50 kWh. As you can see from the data points below, owners tend not to ever go much below 10% SoC, and there is a reason – they have so much capacity. 95.7 kWh according to the ABRP data Model S/X75 Unicorn – BTX8 There are a couple of odd Model S and X 75 with an 85 kWh battery, software limited. It is rumored that they have been fitted with left over BT85 batteries, but the charging curves do not look exactly the same. Anyhow, the result is a battery pack with a lot of extra margin and really fast charging. Lucky owners – 14 of them contribute data to ABRP! TL;DR Battery Code Tesla Model Estimated Usable Capacity 10 kWh -> 50 kWh charge time BT37 3 Long Range 72.8 kWh 23 min BT60 S60 56.3 kWh 42 min BT70 S70 65.7 kWh 33 min BT85 S85 73.4 kWh 27 min BTX4 S/X90 79.8 kWh 23 min BTX5 S/X75 71.6 kWh 27 min BTX6 S/X100 95.7 kWh 20 min BTX8 Rare S/X75 25 min
  23. Bo (ABRP)

    Hyundai Ioniq EV

    Thanks to the 4 Ioniq EV drivers contributing driving data to ABRP, we’ve finally got enough data points to define our first real world consumption curve for the Ioniq EV! We can still use driving data to improve the model even further, so if you’d like to contribute your data, have a look at the instructions. Also, if you’ve got an electric car you’d like us to support have a look at our post detailing How to Add Your Car to ABRP. Much of the data comes from day-to-day driving, with a few road trips sprinkled. Thanks to everyone who contributed their driving data! Ioniq EV Power at Constant Speed The ABRP Model, as you know, is driven by calculating the power consumed for each leg of the trip by plugging in the speed, elevation, and other factors for each segment of the trip, determining how much power the car must output to drive that segment. It then adds up all the segments, and subtracts it from the battery to determine when a charging stop is needed. For this purpose, we need an accurate driving model on flat ground as a baseline, that turns speeds into power. Then we can add all the other driving factors on top of that: The blue dots are consumption samples (30 seconds of driving), adjusted for elevation and speed changes, and the yellow dots are median points within a 1 m/s bin. This means that the red line consumption model is fitted to the median consumption which means that most weather issues, car defects, aggressive driving and so on is typically ignored. As can be seen, we mostly have data points from very low speed and from highway speeds, but that is where things are most interesting. For the Ioniq EV, Hyundai has done a decent job with their aerodynamics, but the big gains come from its small cross-sectional area and low weight. Giving it a reference consumption of only: 7.63 km/kWh (131 Wh/km) at 110 km/h or 4.95 mi/kWh (202 Wh/mile) at 65 mph That’s even lower than the Model 3 (143 Wh/km), which is extremely efficient! The Model 3 has a similar Coefficient of Drag (Cd TM3 = 0.23, Cd Ioniq = 0.24), but a higher cross-sectional area, which increases the overall drag on the car. Weight has less of an impact on range, mainly at lower speeds. At higher speeds, drag is king. We’ve now updated the live model for the Ioniq and removed the “Beta” tag, so you should see the benefits of this higher efficiency in your route planning! Do note that we still set the default a little lower than this measured value, just to ensure we give you a plan that’s not going to over-promise your car’s capabilities. If you know how well your car drives, feel free to re-adjust back to this value! Comparing to the Analytical Model Ioniq Analytical vs Actual driving range. Up to this point, we’ve been using an analytical model, and similar to the Bolt analysis, we see a pretty close fit between the two. Accounting for our standard 10% margin, it’s a pretty solid model! Road Tripping in the Ioniq Given how efficient the Ioniq is, and how quickly it can charge, it might do quite well on a road trip, even given its low range. You’d have to stop regularly to charge, but it might not require as long of a stop. The problem is, that the Ioniq cannot make our standard road trip model. (1000km at 120km/h with 200km legs). Its range is too short. However, shortening the leg distance gives it: Model 3 Long Range: Total trip duration 09:43, of which charging 01:23 Model S100D: Total trip duration 10:05, of which charging 01:45 Model X100D: Total trip duration 10:29, of which charging 02:09 Model S60: Total trip duration 10:35, of which charging 02:15 Ioniq EV: Total trip duration 10:40, of which charging 02:20 (115km legs) Model X60: Total trip duration 11:28, of which charging 03:08 Bolt EV: Total trip duration 12:20, of which charging 04:00 With this adjustment, the road trip requires 8 charging stops, and you’re stopping almost every hour. Given the number of stops, adding “fiddling with the charger” time to that will make your road trip longer, so 10:40 might be an overly optimistic number. Finally, comparing on the charging speed, we get: Ioniq Charge Speed Comparison (metric) Looking at the charge speeds this way, we see an interesting side effect of the small-but-efficient path, as well as an up-side of the more “hockey-stick” shaped charge curves a lot of manufacturers are using in upcoming vehicles. Because the Ioniq is so efficient, it gets a very high relative charge speed, even though it’s actually charging only slightly faster than the Bolt measured by kW. The hockey stick really shows itself when comparing against the S60. Even though it starts out slower, around 50% the Ioniq starts charging faster, and maintains longer than the S60. This is important for the Ioniq’s small battery, and contributes to its decent road tripping time. Appendix: Graphs in Imperial Units
  24. Beautiful, fast new maps! Ioniq! Today we proudly release another update of ABRP. As usual, we continuously update a lot of things under the hood without marking it as new releases, but now we have reached a point where we think it is worth giving it a number: 3.2. News in this release: New maps. With this release we move to MapBox for beautiful, fast maps. It runs noticably smoother in the Tesla browser. The hill shading is particularly interesting for us EV owners. (Real-time traffic display is disabled for the time being as it does not work in the fast version of MapBox. Google Maps is no longer an alternative due to their new pricing polity) Hyundai Ioniq is added as an alpha model. Kona is coming up next, as soon as we get the preliminary model validated. Also worth a read: How to Add Your Car to ABRP. Lots of under-the-hood additions to enable data logging from more cars. Korean charger networks added to allow our friends from the home of Hyundai to plan trips. Enjoy and let us know what you think!

Contact Us

Bo - Lead Developer and Tesla owner: bo@abetterrouteplanner.com

Jason - New Car Models, Developer and Bolt owner : jason@abetterrouteplanner.com

Idreams - Forums Administrator, Forums Developer and Tesla owner : idreams@abetterrouteplanner.com

×
×
  • Create New...